A method and system for inverting vegetation ecological coverage based on remote sensing data

Abstract

The invention discloses a vegetation ecological coverage inversion method and system based on remote sensing data, and relates to the field of remote sensing data. Based on big data, multi-source acquisition time-remote sensing data corresponding to each preset target area is obtained, plants in the target area are classified according to the acquisition time-remote sensing data, plant communities are obtained, and then the distribution states corresponding to the plant communities are obtained, a coverage recognition model is constructed, and based on the coverage recognition model, distribution states-coverages corresponding to different distribution states of plant communities are obtained, and a coverage law data tree is generated. The method for constructing the coverage recognition model includes: based on big data, a plurality of coverages corresponding to the corresponding plant community-distribution states of each acquisition time-plant community are obtained, marked as distribution state-sub-coverages, and then the obtained distribution state-sub-coverages are used to generate a distribution state-sub-coverage data set, and then a discrete function is generated from the distribution state-sub-coverage data set.

Description

Technical Field

[0001] The invention relates to the field of remote sensing data, and in particular to a vegetation ecological coverage inversion method and system based on remote sensing data. Background Art

[0002] Vegetation ecological coverage refers to the percentage of the vertical projection area of ​​vegetation (including leaves, stems, and branches) on the ground to the total area of ​​the statistical area; vegetation coverage can be divided into four types: high, medium-high, medium, and low. Vegetation coverage is an important indicator for measuring the status of surface vegetation, an important basic data for describing ecosystems, and an important indicator of regional ecosystem environmental changes. It is of great significance to hydrology, ecology, regional changes, etc.

[0003] Under normal circumstances, coverage is mainly detected by surface measurement and remote sensing monitoring. However, during the use of this method, due to the diversity of vegetation ecology, as well as changes in land use and land cover, the coverage is affected. In addition, due to the limitations on the accuracy of remote sensing data, there are certain coverage errors, resulting in inaccurate remote sensing data. Remote sensing data usually has periodicity, and there are certain difficulties in measuring the coverage of vegetation ecology that needs to be monitored in real time, and real-time monitoring and evaluation are impossible. Therefore, a vegetation ecological coverage inversion method and system based on remote sensing data is provided. Summary of the invention

[0004] In order to solve the above technical problems, the present invention provides a vegetation ecological coverage inversion method and system based on remote sensing data;

[0005] The purpose of the present invention can be achieved by the following technical solution: a vegetation ecological coverage inversion method based on remote sensing data, the method comprising the following steps:

[0006] Step S1: Set a number of target areas and obtain the corresponding acquisition time-remote sensing data of each target area based on big data multi-source;

[0007] Step S2: Classify the plants in the target area according to the acquisition time-remote sensing data, obtain the plant community, and then obtain the distribution state corresponding to the plant community, build a coverage recognition model, and based on the coverage recognition model, obtain the distribution state-coverage corresponding to different distribution states of the plant community, and generate a coverage law data tree;

[0008] Step S3: setting a number of new target areas and collecting corresponding remote sensing data, and generating a coverage training sample data set based on the remote sensing data;

[0009] Step S4: Send the coverage training sample data set to the coverage law data tree, perform sample data training on the coverage law data tree, and obtain an accurate coverage law data tree;

[0010] Step S5: construct a collection time-distribution status-coverage inversion model based on the precise coverage law data tree to obtain the coverage corresponding to the target area.

[0011] Furthermore, the acquisition process of the acquisition time-remote sensing data includes:

[0012] The acquisition time corresponding to the remote sensing data is obtained based on big data for each target area, and the remote sensing data corresponding to the same acquisition time is extracted, marked as acquisition time-remote sensing data, and generated as the first data tree branch.

[0013] Furthermore, the process of obtaining the plant community includes:

[0014] Obtain the color tone of the acquisition time-remote sensing data, classify the plants in the target area according to the different color tones of the acquisition time-remote sensing data, divide the areas with the same color tone into the same plant community, and then generate several plant communities;

[0015] Tagging the collection time and the plant community to generate collection time-plant community, and then connecting several collection time-plant community and collection time-remote sensing data to generate a second data tree branch;

[0016] According to the acquisition time-remote sensing data, the distribution state corresponding to the acquisition time-plant community is obtained and marked as plant community-distribution state;

[0017] The distribution state includes spectral characteristics, geometric shape and image size.

[0018] Furthermore, the process of constructing a coverage recognition model to obtain a distribution state-coverage includes:

[0019] Based on big data, several coverages corresponding to each collection time-plant community-distribution state are obtained, which are marked as distribution state-sub-coverage;

[0020] The obtained distribution state-sub-coverage is used to generate a distribution state-sub-coverage data set, and the maximum and minimum values ​​of the distribution state-sub-coverage are eliminated. Then, the distribution state-sub-coverage data set is used to generate a discrete function, and the discrete degree corresponding to the distribution state-sub-coverage is obtained based on the discrete function. That is, the formula for obtaining the discrete function is:

[0021] ;

[0022] in, It is represented as the discrete degree corresponding to the distribution state-sub-coverage, i represents the ith distribution state-sub-coverage in the distribution state-sub-coverage data set, x, y and z represent the spectral characteristics, geometric shape and image size corresponding to the distribution state, respectively. It is expressed as the distribution state-subcoverage corresponding to the i-th distribution state x, y, z, It is represented as the average of the distribution state-sub-coverage data set, and N is represented as the total number of distribution state-sub-coverages in the distribution state-sub-coverage data set; where i=1, 2,..., N, and N is a positive integer.

[0023] Furthermore, the discrete degree threshold is set to , compare the discreteness corresponding to the distribution state-sub-coverage with the discreteness threshold;

[0024] Will The corresponding distribution state-sub-coverage is eliminated, and the The corresponding distribution state-sub-coverage is retained in the distribution state-sub-coverage data set, and the maximum and minimum values ​​of the distribution state-sub-coverage in the distribution state-sub-coverage data set are obtained and marked as and , and then obtain the distribution state - coverage, that is, the formula is:

[0025] ;

[0026] in, It is expressed as the distribution state-coverage corresponding to the plant community-distribution state at the collection time t, It is expressed as the average of the distribution state-subcoverage data retained in the distribution state-subcoverage data set;

[0027] The plant community-distribution state and the corresponding distribution state-coverage data are connected to generate a third data tree branch.

[0028] Furthermore, the generation process of the coverage regularity data tree includes:

[0029] Each target area is processed through step S2 to generate a corresponding sub-coverage law data tree, and the sub-coverage law data trees corresponding to each target area are connected with the target area as a branching node to generate a coverage law data tree.

[0030] The process of obtaining the corresponding coverage of the target area includes:

[0031] If the remote sensing data of the target area are acquired at the same acquisition time, then based on the coverage law data tree, the acquisition time - distribution state corresponding to each plant community in the plant community - distribution state - coverage is added to obtain the coverage corresponding to the acquisition time - remote sensing data, and generate the target area - acquisition time - coverage;

[0032] If the remote sensing data corresponding to each plant community in the target area is obtained according to different distribution states at different collection times, the collection time-plant community is obtained based on the coverage law data tree according to the different collection times, and then the plant community-distribution state is obtained according to the different distribution states, and then the distribution state-coverage corresponding to the plant community-distribution state is obtained, and the obtained distribution state-coverage is added to obtain the target area-coverage.

[0033] Furthermore, the generation process of the coverage training sample data includes:

[0034] Set a new collection time and collect the remote sensing data corresponding to several plant communities in the new target area at the same time, obtain the corresponding distribution status according to the remote sensing data corresponding to each plant community, and mark it as new collection time-new distribution status;

[0035] By conducting field surveys on the plant communities corresponding to the new collection time and new distribution status, the actual coverage is obtained, and then the actual coverage corresponding to each plant community in the new target area is added together to generate a new coverage corresponding to the new target area;

[0036] Generate the first coverage training sample data according to the new collection time, new distribution state and new coverage;

[0037] Based on big data, we obtain various plant communities in the new target area, set the corresponding community collection time, obtain the remote sensing data corresponding to each plant community according to the community collection time, and then obtain the community distribution status corresponding to each plant community. We conduct actual investigation based on the remote sensing data corresponding to the plant community to obtain the community coverage corresponding to each plant community.

[0038] The coverage training sample data is generated according to the community distribution status and community distribution time, and marked as (community collection time, community distribution status, community coverage), and then the coverage training sample data corresponding to several new target areas are stored to generate a coverage training sample data set.

[0039] Furthermore, the process of obtaining the precise coverage regularity data tree includes:

[0040] Send the community collection time and community distribution state in the coverage training sample data set to the coverage law data tree, traverse the distribution state corresponding to the collection time in the coverage law data tree, obtain the collection time corresponding to the community collection time, and then obtain the distribution state corresponding to the community distribution state, and then obtain the distribution state corresponding to the coverage law data tree - coverage;

[0041] According to the distribution status - coverage and community coverage, the plant community coverage deviation value is obtained, that is, the formula is:

[0042]

[0043] in, It is expressed as the plant community coverage deviation value, is represented as the new coverage; ;

[0044] The plant community coverage deviation value is compared with the preset plant community coverage deviation value threshold range. Make comparisons;

[0045] like , then the corresponding distribution state - coverage will be retained;

[0046] like , then the corresponding distribution state-coverage is replaced by community coverage;

[0047] After all the coverage training sample data in the coverage training sample data set are trained, an accurate coverage rule data tree is generated.

[0048] Furthermore, the process of obtaining the coverage corresponding to the target area includes:

[0049] Based on big data, remote sensing data corresponding to the target area is obtained, and each plant community in the target area is obtained according to the remote sensing data. The corresponding collection time is set for each plant community, and the remote sensing data corresponding to the plant community is obtained according to the collection time. Then, the distribution status corresponding to the plant community is obtained, and the collection time-distribution status is generated. The collection time-distribution status is sent to the collection time-distribution status-coverage inversion model, and the distribution status-coverage corresponding to each plant community in the target area is obtained and added to obtain the coverage corresponding to the target area.

[0050] Furthermore, the system includes the following modules:

[0051] Acquisition module: used to obtain the acquisition time corresponding to each target area based on big data and multiple sources - remote sensing data;

[0052] Generation module: used to build a coverage recognition model based on the acquisition time-remote sensing data, and then generate a coverage law data tree;

[0053] Sample module: used to set several new target areas and collect corresponding remote sensing data, and generate coverage training sample data sets based on remote sensing data;

[0054] Training module: used to perform sample data training on the coverage law data tree according to the coverage training sample data set to obtain an accurate coverage law data tree;

[0055] Inversion model module: used to build an acquisition time-distribution status-coverage inversion model based on the precise coverage law data tree to obtain the corresponding coverage of the target area.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] 1. The present invention obtains remote sensing data of a target area based on big data, obtains several plant communities in the target area according to the remote sensing data, obtains corresponding coverage according to different collection times and different distribution states, and generates a coverage training sample data set; trains the coverage training sample data set to obtain an accurate coverage law data tree, and constructs an acquisition time-distribution state-coverage inversion model based on the accurate coverage law data tree to obtain the corresponding coverage of the target area; usually, the remote sensing data of the target area is obtained at the same time, and then the remote sensing data is processed and analyzed to obtain the coverage of the target area, but the distribution state of the target area and the influence of the collection time may cause coverage errors, so the present invention improves the accuracy of coverage;

[0058] 2. The present invention divides the target area into plant communities and sets real-time collection time to achieve real-time monitoring of plant communities, obtain corresponding distribution states, and then obtain the corresponding distribution states of plant communities - coverage, thereby achieving real-time monitoring of coverage in the target area. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0060] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0061] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0062] like Figure 1 As shown, a vegetation ecological coverage inversion method based on remote sensing data comprises the following steps:

[0063] Step S1: Set a number of target areas and obtain the corresponding acquisition time-remote sensing data of each target area based on big data multi-source;

[0064] The remote sensing data include aerial photos and satellite photos;

[0065] For each target area, the acquisition time corresponding to the remote sensing data is obtained based on the big data, and the remote sensing data corresponding to the same acquisition time is extracted, marked as acquisition time-remote sensing data, and generated as the first data tree branch;

[0066] It should be further explained that, in the specific implementation process, a target area is set, and the target area may be one or more.

[0067] Step S2: Classify the plants in the target area according to the acquisition time-remote sensing data, obtain the plant community, and then obtain the distribution state corresponding to the plant community, build a coverage recognition model, and based on the coverage recognition model, obtain the distribution state-coverage corresponding to different distribution states of the plant community, and generate a coverage law data tree, and then obtain the coverage corresponding to the target area;

[0068] The target area is composed of one or more plant communities;

[0069] Obtain the color tone of the acquisition time-remote sensing data, classify the plants in the target area according to the different color tones of the acquisition time-remote sensing data, divide the areas with the same color tone into the same plant community, complete the division of the target area, and generate several plant communities;

[0070] Tagging the collection time and the plant community to generate collection time-plant community, and then connecting several collection time-plant community and collection time-remote sensing data to generate a second data tree branch;

[0071] According to the acquisition time-remote sensing data, the distribution state corresponding to the acquisition time-plant community is obtained and marked as plant community-distribution state;

[0072] The distribution state includes spectral characteristics, geometric shape and image size.

[0073] It should be further explained that, in the specific implementation process, at the same collection time, different plant communities have different corresponding distribution states, and thus different distribution states lead to different obtained coverages. Therefore, according to different distribution states, the coverages corresponding to several collection times and plant communities are obtained. In a specific embodiment, due to the diversity of vegetation ecology, the present invention improves the accuracy of the coverages obtained for each plant community under different distribution states.

[0074] The process of constructing the coverage recognition model to obtain the distribution state-coverage includes:

[0075] Based on big data, several coverages corresponding to the plant community-distribution state at each collection time-plant community are obtained, marked as distribution state-sub-coverages, and then the distribution state-coverage corresponding to the plant community-distribution state is obtained according to the distribution state-sub-coverages;

[0076] The obtained distribution state-sub-coverage is used to generate a distribution state-sub-coverage data set, the maximum and minimum values ​​of the distribution state-sub-coverage are eliminated, and then a discrete function is generated from the distribution state-sub-coverage data set, and the discrete degree corresponding to the distribution state-sub-coverage is obtained based on the discrete function;

[0077] That is, the specific formula is:

[0078]

[0079] in, It is represented as the discrete degree corresponding to the distribution state-sub-coverage, i represents the ith distribution state-sub-coverage in the distribution state-sub-coverage data set, x, y and z represent the spectral characteristics, geometric shape and image size corresponding to the distribution state, respectively. It is expressed as the distribution state-subcoverage corresponding to the i-th distribution state x, y, z, It is represented as the average of the distribution state-subcoverage data set, and N is represented as the total number of distribution state-subcoverages in the distribution state-subcoverage data set; wherein, i=1, 2, ..., N, and N is a positive integer;

[0080] Set the discreteness threshold to , compare the discreteness corresponding to the distribution state-sub-coverage with the discreteness threshold;

[0081] Will The corresponding distribution state-sub-coverage is eliminated, and the The corresponding distribution state-sub-coverage is retained in the distribution state-sub-coverage data set, and then the maximum and minimum values ​​of the distribution state-sub-coverage in the distribution state-sub-coverage data set are obtained, which are marked as and , and then obtain the distribution state - coverage; the specific formula is:

[0082]

[0083] in, It is expressed as the distribution state corresponding to the plant community-distribution state at the collection time t-cover;

[0084] The plant community-distribution state and the corresponding distribution state-coverage data are connected to generate a third data tree branch.

[0085] The generation process of the coverage regularity data tree includes:

[0086] The target area is taken as the trunk, the collection time is taken as the first bifurcation node of the trunk, the first data branch is connected with the first bifurcation node, the plant community is taken as the second bifurcation node of the first data branch, the second data branch is connected with the second bifurcation node, the distribution state is taken as the third bifurcation node of the second data branch, the third data branch is connected with the third bifurcation node, and a sub-coverage regularity data tree is generated;

[0087] Each target area is processed through step S2 to generate a corresponding sub-coverage law data tree, and the sub-coverage law data trees corresponding to each target area are connected with the target area as a bifurcation node to generate an accurate coverage law data tree.

[0088] The process of obtaining the coverage corresponding to the target area includes:

[0089] If the remote sensing data of the target area are acquired at the same acquisition time, then based on the coverage law data tree, the acquisition time - distribution state corresponding to each plant community in the plant community - distribution state - coverage is added to obtain the coverage corresponding to the acquisition time - remote sensing data, and generate the target area - acquisition time - coverage;

[0090] If the remote sensing data corresponding to each plant community in the target area is obtained according to different distribution states at different collection times, the collection time-plant community is obtained based on the coverage law data tree according to the different collection times, and then the plant community-distribution state is obtained according to the different distribution states, and then the distribution state-coverage corresponding to the plant community-distribution state is obtained, and the obtained distribution state-coverage is added to obtain the target area-coverage;

[0091] The coverage includes target area-collection time-coverage and target area-coverage.

[0092] It should be further explained that, in the specific implementation process, the coverage of the target area under different collection time and different distribution states is obtained; in the prior art, the remote sensing data of the target area is usually obtained at the same time, and then the remote sensing data is processed and analyzed to obtain the coverage of the target area. Because there are several plant communities in the target area, each plant community has a different distribution state. If several plant communities in the target area are collected at the same time, there may be errors in coverage depending on the distribution state. Therefore, the present invention obtains the corresponding coverage of several plant communities in the target area according to different collection times and different distribution states, and then obtains the coverage of the target area, thereby improving the accuracy of the coverage.

[0093] Step S3: setting a number of new target areas and collecting corresponding remote sensing data, and generating a coverage training sample data set based on the remote sensing data;

[0094] The generation process of the coverage training sample data includes:

[0095] Set a new collection time and collect the remote sensing data corresponding to several plant communities in the new target area at the same time, obtain the corresponding distribution status according to the remote sensing data corresponding to each plant community, and mark it as new collection time-new distribution status;

[0096] By conducting field surveys on the plant communities corresponding to the new collection time and new distribution status, the actual coverage is obtained, and then the actual coverage corresponding to each plant community in the new target area is added together to generate a new coverage corresponding to the new target area;

[0097] Generate the first coverage training sample data according to the new collection time - new distribution state and new coverage, marked as (new collection time - new distribution state, new coverage);

[0098] In a specific embodiment, two different coverage training sample data are set to perform data training on the coverage rule data tree, thereby avoiding the particularity of the data, reducing the contingency, and thus improving the accuracy;

[0099] Based on big data, we obtain various plant communities in the new target area, set the corresponding community collection time, obtain the remote sensing data corresponding to each plant community according to the community collection time, and then obtain the community distribution status corresponding to each plant community. We conduct actual investigation based on the remote sensing data corresponding to the plant community to obtain the community coverage corresponding to each plant community.

[0100] Generate coverage training sample data according to the community distribution status and community distribution time, marked as (community collection time, community distribution status, community coverage);

[0101] The coverage training sample data corresponding to several new target areas are stored to generate a coverage training sample data set.

[0102] Step S4: Send the coverage training sample data set to the coverage law data tree, perform sample data training on the coverage law data tree, and obtain an accurate coverage law data tree;

[0103] The process of obtaining the precise coverage regularity data tree includes:

[0104] Send the community collection time and community distribution state in the coverage training sample data set to the coverage law data tree, traverse the distribution state corresponding to the collection time in the coverage law data tree, obtain the collection time corresponding to the community collection time, and then obtain the distribution state corresponding to the community distribution state, and then obtain the distribution state corresponding to the coverage law data tree - coverage;

[0105] According to the distribution status - coverage and community coverage, the plant community coverage deviation value is obtained;

[0106] That is, the specific formula is:

[0107]

[0108] in, It is expressed as the plant community coverage deviation value, is represented as the new coverage; ;

[0109] The plant community coverage deviation value is compared with the preset plant community coverage deviation value threshold range. Make comparisons;

[0110] like , then the corresponding distribution state - coverage will be retained;

[0111] like , then the corresponding distribution state-coverage is replaced by community coverage;

[0112] After all the coverage training sample data in the coverage training sample data set are trained, an accurate coverage law data tree is generated;

[0113] In a specific embodiment, the first coverage training sample data is sent to the coverage law data tree, the distribution state-coverage corresponding to the plant community at the same collection time point is obtained, and the corresponding plant community coverage deviation value is obtained, and compared with the plant community coverage deviation value corresponding to the plant community obtained at different collection times, and the coverage law data tree is further confirmed;

[0114] It should be further explained that, in the specific implementation process, the coverage training sample data set is infinite. The coverage training sample data in the coverage training sample data set can not only be obtained through step S4, but also can be obtained through visual estimation, sampling, instrumental method and model method; the coverage law data tree is trained and optimized through the acquired coverage training sample data set, thereby improving the accuracy of obtaining coverage.

[0115] Step S5: constructing a collection time-distribution state-coverage inversion model based on the precise coverage law data tree to obtain the coverage corresponding to the target area;

[0116] Based on big data, remote sensing data corresponding to the target area is obtained. According to the remote sensing data, each plant community in the target area is obtained. The corresponding collection time is set for each plant community. According to the collection time, the remote sensing data corresponding to the plant community is obtained, and then the distribution state corresponding to the plant community is obtained. The collection time-distribution state is generated, and the collection time-distribution state is sent to the collection time-distribution state-coverage inversion model, and the distribution state-coverage corresponding to each plant community in the target area is obtained and added to obtain the coverage corresponding to the target area;

[0117] It should be further explained that, in the specific implementation process, the target area is divided into plant communities, which can be monitored in real time to obtain the corresponding distribution status, and then the corresponding distribution status of the plant community - coverage - is obtained, thus realizing real-time monitoring of the coverage of the target area.

[0118] The present invention also discloses an inversion system of a vegetation ecological coverage inversion method based on remote sensing data, and the system comprises the following modules:

[0119] Acquisition module: used to obtain the acquisition time corresponding to each target area based on big data and multiple sources - remote sensing data;

[0120] Generation module: used to build a coverage recognition model based on the acquisition time-remote sensing data, and then generate a coverage law data tree;

[0121] Sample module: used to set several new target areas and collect corresponding remote sensing data, and generate coverage training sample data sets based on remote sensing data;

[0122] Training module: used to perform sample data training on the coverage law data tree according to the coverage training sample data set to obtain an accurate coverage law data tree;

[0123] Inversion model module: used to build an acquisition time-distribution status-coverage inversion model based on the precise coverage law data tree to obtain the corresponding coverage of the target area.

[0124] The features and exemplary embodiments of various aspects of the present application are described in detail above. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is further described in detail in combination with the drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application rather than to limit the present application. For those skilled in the art, the present application can be implemented without some of these specific details. The above description of the embodiments is merely to provide a better understanding of the present application by showing examples of the present application.

[0125] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A vegetation ecological coverage inversion method based on remote sensing data, characterized in that: The method comprises the following steps: Step S1: Set a number of target areas and obtain the corresponding acquisition time-remote sensing data of each target area based on big data multi-source; Step S2: Classify the plants in the target area according to the acquisition time-remote sensing data, obtain the plant community, and then obtain the distribution state corresponding to the plant community, build a coverage recognition model, and based on the coverage recognition model, obtain the distribution state-coverage corresponding to different distribution states of the plant community, and generate a coverage law data tree; The method for constructing a coverage identification model includes: obtaining a number of coverages corresponding to each collection time-plant community in the corresponding plant community-distribution state based on big data, marking them as distribution state-sub-coverages, and then generating a distribution state-sub-coverage data set from the obtained distribution state-sub-coverages, and then generating a discrete function from the distribution state-sub-coverage data set, and obtaining the discrete degree corresponding to the distribution state-sub-coverage based on the discrete function; The method for obtaining the distribution state-coverage corresponding to different distribution states of plant communities is as follows: set the discrete degree threshold to , the discrete degree corresponding to the distribution state-sub-coverage Compare with the discreteness threshold; Will The corresponding distribution state-sub-coverage is eliminated, and the The corresponding distribution state-sub-coverage is retained in the distribution state-sub-coverage data set, and the maximum and minimum values ​​of the distribution state-sub-coverage in the distribution state-sub-coverage data set are obtained and marked as and , and then obtain the distribution state - coverage, that is, the formula is: ； in, It is expressed as the distribution state-coverage corresponding to the plant community-distribution state at the collection time t, It is expressed as the average of the distribution state-subcoverage data retained in the distribution state-subcoverage data set; Connect the data of plant community-distribution state and corresponding distribution state-coverage to generate a third data tree branch; Step S3: setting a number of new target areas and collecting corresponding remote sensing data, and generating a coverage training sample data set based on the remote sensing data; Step S4: Send the coverage training sample data set to the coverage law data tree, perform sample data training on the coverage law data tree, and obtain an accurate coverage law data tree; Step S5: construct a collection time-distribution status-coverage inversion model based on the precise coverage law data tree to obtain the coverage corresponding to the target area.

2. The vegetation ecological coverage inversion method based on remote sensing data according to claim 1 is characterized in that: The acquisition process of the acquisition time-remote sensing data includes: The acquisition time corresponding to the remote sensing data is obtained based on big data for each target area, and the remote sensing data corresponding to the same acquisition time is extracted, marked as acquisition time-remote sensing data, and generated as the first data tree branch.

3. The vegetation ecological coverage inversion method based on remote sensing data according to claim 2 is characterized in that: The process of obtaining the plant community includes: Obtain the color tone of the acquisition time-remote sensing data, classify the plants in the target area according to the different color tones of the acquisition time-remote sensing data, divide the areas with the same color tone into the same plant community, and then generate several plant communities; Tagging the collection time and the plant community to generate collection time-plant community, and then connecting several collection time-plant community and collection time-remote sensing data to generate a second data tree branch; According to the acquisition time-remote sensing data, the distribution state corresponding to the acquisition time-plant community is obtained and marked as plant community-distribution state; The distribution state includes spectral characteristics, geometric shape and image size.

4. The vegetation ecological coverage inversion method based on remote sensing data according to claim 3 is characterized in that: The process of constructing the coverage recognition model to obtain the distribution state-coverage includes: The formula for obtaining the discrete degree corresponding to the distribution state-sub-coverage is: ； in, It is represented as the discrete degree corresponding to the distribution state-sub-coverage, i represents the ith distribution state-sub-coverage in the distribution state-sub-coverage data set, x, y and z represent the spectral characteristics, geometric shape and image size corresponding to the distribution state, respectively. It is expressed as the distribution state-subcoverage corresponding to the i-th distribution state x, y, z, It is represented as the average of the distribution state-sub-coverage data set, and N is represented as the total number of distribution state-sub-coverages in the distribution state-sub-coverage data set; where i=1, 2,..., N, and N is a positive integer.

5. The vegetation ecological coverage inversion method based on remote sensing data according to claim 4 is characterized in that: The generation process of the coverage regularity data tree includes: Each target area is processed through step S2 to generate a corresponding sub-coverage law data tree, and the sub-coverage law data trees corresponding to each target area are connected with the target area as a branching node to generate a coverage law data tree.

6. The vegetation ecological coverage inversion method based on remote sensing data according to claim 5 is characterized in that: The generation process of the coverage training sample data includes: Based on big data, we obtain various plant communities in the new target area, set the corresponding community collection time, obtain the remote sensing data corresponding to each plant community according to the community collection time, and then obtain the community distribution status corresponding to each plant community. We conduct actual investigation based on the remote sensing data corresponding to the plant community to obtain the community coverage corresponding to each plant community. The coverage training sample data is generated according to the community distribution status and community distribution time, and marked as (community collection time, community distribution status, community coverage), and then the coverage training sample data corresponding to several new target areas are stored to generate a coverage training sample data set.

7. The vegetation ecological coverage inversion method based on remote sensing data according to claim 6 is characterized in that: The process of obtaining the precise coverage regularity data tree includes: Send the community collection time and community distribution state in the coverage training sample data set to the coverage law data tree, traverse the distribution state corresponding to the collection time in the coverage law data tree, obtain the collection time corresponding to the community collection time, and then obtain the distribution state corresponding to the community distribution state, and then obtain the distribution state corresponding to the coverage law data tree - coverage; According to the distribution status - coverage and community coverage, the plant community coverage deviation value is obtained, that is, the formula is: in, It is expressed as the plant community coverage deviation value, is represented as the new coverage; ; The plant community coverage deviation value is compared with the preset plant community coverage deviation value threshold range. Make comparisons; like , then the corresponding distribution state - coverage will be retained; like , then the corresponding distribution state-coverage is replaced by community coverage; After all the coverage training sample data in the coverage training sample data set are trained, an accurate coverage rule data tree is generated.

8. The vegetation ecological coverage inversion method based on remote sensing data according to claim 7 is characterized in that: The process of obtaining the coverage corresponding to the target area includes: Based on big data, remote sensing data corresponding to the target area is obtained, and each plant community in the target area is obtained according to the remote sensing data. The corresponding collection time is set for each plant community, and the remote sensing data corresponding to the plant community is obtained according to the collection time. Then, the distribution status corresponding to the plant community is obtained, and the collection time-distribution status is generated. The collection time-distribution status is sent to the collection time-distribution status-coverage inversion model, and the distribution status-coverage corresponding to each plant community in the target area is obtained and added to obtain the coverage corresponding to the target area.

9. An inversion system for the vegetation ecological coverage inversion method based on remote sensing data according to any one of claims 1 to 8, characterized in that: The system includes the following modules: Acquisition module: used to obtain the acquisition time-remote sensing data corresponding to each target area based on big data and multiple sources; Generation module: used to build a coverage recognition model based on the acquisition time-remote sensing data, and then generate a coverage law data tree; Sample module: used to set several new target areas and collect corresponding remote sensing data, and generate coverage training sample data sets based on remote sensing data; Training module: used to perform sample data training on the coverage law data tree according to the coverage training sample data set to obtain an accurate coverage law data tree; Inversion model module: used to construct the acquisition time-distribution status-coverage inversion model based on the precise coverage law data tree to obtain the corresponding coverage of the target area.

Images